This paper analyses the role of a public development bank when banks use a costly screening technology to make credit decisions. We explore two issues: 1) which types of firms should be optimally targeted by public financial support; and 2) what type of mechanism should be implemented in order to efficiently support the targeted firms’ access to credit. We show that, in the presence of costly screening, the market leads to an inefficient allocation, as there will be underprovision of credit. The market imperfection results from the inability of banks to appropriate the full benefits of projects they finance. This implies that the misallocation of credit is more pronounced for high value projects. This central result, and its implication that PDBs could play a central role in the financing of high value projects, contrast with the usual emphasis on credit underprovision for relatively weak projects/firms (SMEs, young firms, those without collateral, etc.). We show that a public development bank may alleviate the inefficiencies by lending to commercial banks at subsidized rates and targeting the firms that generate high added value. This may be implemented through subsidized ear-marked lending to the banks or through credit guarantees which, in “normal times”, we show to be equivalent. Still, when banks are facing a liquidity shortage, lending is preferred, while when banks are undercapitalized, a credit guarantees program is best suited to alleviate the constraints banks’ face.